{
  "abstract": "Solving large-scale multiparametric quadratic programs (mpQPs) in real time often requires thousands of iterative optimization updates, creating a major computational bottleneck in model predictive control and learning-enabled systems. This paper proposes TQPNet, a self-supervised neural compression framework that learns to compress long ADMM optimization trajectories into a compact neural warm start followed by a few refinement iterations. We derive a low-dimensional ADMM scheme operating on the primal-dual variables $(x,β)$ and show that its iteration admits an equivalent ReLU-layer representation, termed ADMM(ReLU). Building on this insight, TQPNet combines a multilayer perceptron with ReLU activations (MLP(ReLU)) and a small number of unrolled ADMM(ReLU) refinement layers. The network is trained offline using an optimality-condition-informed residual loss, eliminating the need for solver-generated labels. The proposed residual-based framework enables both self-supervised training and online solution certification through adaptive ADMM(ReLU) refinement. Experiments on large-scale mpQPs and real-time MPC applications demonstrate that TQPNet achieves high accuracy, strong generalization, and substantially faster inference than conventional optimization-based methods.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.38079",
  "author_total": 6,
  "authors": [
    {
      "name": "Liang Wu"
    },
    {
      "name": "Honghui Zheng"
    },
    {
      "name": "Bo Yang"
    },
    {
      "name": "Xu Yang"
    },
    {
      "name": "Yilin Mo"
    },
    {
      "name": "Ján Drgoňa"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/9kzhscd35jj5r.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.38079",
  "fulltext_url": null,
  "id": "9kzhscd35jj5r",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.38079",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "cs.SY"
    ],
    [
      "arxiv",
      "eess.SY"
    ],
    [
      "arxiv",
      "math.OC"
    ]
  ],
  "openalex": null,
  "pdf_url": "https://arxiv.org/pdf/2610.38079",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-27",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/9kzhscd35jj5r.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:36:32Z",
      "native_id": "2610.38079",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.38079"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:math.OC",
      "subfield": "optimisation-control"
    }
  ],
  "title": "Neural Compression of Long ADMM Trajectory for Multiparametric Quadratic Program",
  "title_text": "Neural Compression of Long ADMM Trajectory for Multiparametric Quadratic Program",
  "type": "preprint",
  "updated_at": "2026-11-01T12:20:50Z",
  "url": "https://research.draconitic.com/paper/9kzhscd35jj5r",
  "venue": "arXiv",
  "versions": []
}